|
| Greg's diary |
| December 2005 |
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Thursday, 1 December 2005 | Carnate (Milano) –> Siena –> Agliana | Images for 1 December 2005 |
| Top of page | ||
| next day | ||
| last day |
Up early again with the intention of making it to Siena by lunch time. Many things got in the way: first, there had been quite a frost, and of course the car didn't contain an ice scraper (nor a yellow vest for emergencies, which is apparently required by law), and the windscreen washer didn't work. At the last minute, George Georgalis had decided to come, so into Carnate to pick him up, and ended up waiting in front of a level crossing in Carnate, which looked like it would take some time. Saw a lot of people turning left, so followed them, assuming they knew a way across the railway lines; but they were all going elsewhere, and I got myself thoroughly lost in a maze of twisty little roads, all different.
Finally found my way back (in fact, after less than 10 minutes), and off into the traffic jam that was the road to the Tangenziali. I had intended to pick Russell up at about 8:30, but we arrived a whole hour late.
Down the Po valley with less than perfect weather conditions: fog and later snow, coupled with some road works from hell. Across the mountains to Toscana, where the difference couldn't have been more marked: sunshine and warmth. Dropped George near Firenze Nord and continued to Siena, missing the fast road in the process. I hate bad maps! Couldn't find any information on where the correct road was, so along the SS 222, having lunch in Greve in Chianti.
By the time we got to Siena, it was 16:00—I had intended to be there at 13:00—and we barely had time to rush through the Pizza del Campo and out again. That was worthwhile, however: it seems that all of the Contrade had sent a delegation of two flag wavers and a drummer, which each disappeared into the Palazzo Pubblico:
We weren't given details of how to get to the return location (the main Firenze location), so on the way to Siena, Russell tried calling them. They didn't answer. Called the AVIS main enquiry line, who suggested returning to the airport, which is supposed to be open until 11 pm.
I wish I understood this more: is this a daily event, or were we lucky?
Back to Firenze to drop the car, not without difficulties. We found the fast road on the way back, but forgot to fill up, and had to leave again and fill up at Colle de Val d'Elsa, where Doris and I had stayed nearly 30 years ago. To add to its other problems, it's almost impossible to fill the tank of this car. The last 15 litres (of 40) went in at a snail's pace. Also had the devil's own job finding the airport, and when we got there there was nobody to accept the car. Not happy.
Met up with Gianmarco Giovanelli there, and off to Prato to find a hotel. Ended up at the Albergo il Villino in Agliana, not expensive but adequate, and with the interesting advantage of free Internet connections in every room. Then into Prato for a pleasant Tuscan style meal at a little Osteria in the Piazza Mercatale, and late to bed.
| Friday, 2 December 2005 | Agliana –> Firenze –> Pisa –> Perugia | Images for 2 December 2005 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Off this morning by train with George and Russell to Firenze to see the sights. They decided to go their own way: I was “rushing” them too much, not surprising in view of the fact that we had just shy of 2 hours to spend. I managed it as far as the Piazza Signoria:
|
|||||||||||||||||||||||
and the Ponte Vecchio:
|
|||||||||||||||||||||||
Was just returning to the station when I got a call from George saying “Don't catch that train!” (the 12:55 that Gianmarco had asked us to catch), and blaming me for “changing” the agreement we had had with Gianmarco. Then the connection was dropped: the phone (Russell's) was out of credit. George's has been that way for over a week.
Called up Gianmarco and discussed the matter, then called George back (his phone still accepts incoming calls; Russell's didn't) and agreed with him that they'd come back to the station for the next train, which left in 15 minutes at 13:19. It takes about 8 minutes from where they were (near the cathedral, as far as they went) to the station, so that should have been sufficient. After 10 minutes, called, and they said they were a block away. Another 5 minutes, and they still hadn't shown. Called again and discovered that George had decided to go shopping. He knew that there would be another train in another 20 minutes (there wasn't). Furious, left the job to Gianmarco, who agreed with them that they would be on the 14:00 train. Caught that without seeing the others. Arrived in Agliana to find Gianmarco talking with them on the phone: they had got on the wrong train and got off at Montecatini Terme, which had absolutely nothing to do with where they should have gone. We spent some time scratching our heads about how they managed to make such a mistake. Maybe it has something to do with George's knowledge that there was a train round 13:40.
Down there to pick them up, then on to Pisa, where we had a metered 15 minutes to see the miracles:
|
|||||||||||||||||||||||
After that, Gianmarco dropped me off with Luigi Rizzo, with whom I drove first back to Prato to pick up some other people, then on to a somewhat wet Perugia, where they had arranged dinner at a place chosen apparently mainly for the name: Osteria il Gufo. English speakers may think of Walt Disney when they hear the name Gufi, but here it's the plural of Gufo, which is a kind of owl. Food was very good, though, and far too late back to the hotel, getting soaked in the rain in the process.
| Saturday, 3 December 2005 | Perugia | Images for 3 December 2005 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
The GufiCon started today. Downstairs at 8:30 as requested to discover nobody around. Had breakfast and admired the pouring rain; after about 45 minutes decided that nobody would come, so enquired where the university was and out in still pouring rain to find it. The instructions were pretty vague, and ended up outside the city walls before giving up. On the way back had the double coincidence of encountering the person who had given me the instructions as we were both outside the venue, which proved to be a 12th century abbey attached to the Church of San Pietro. Inside, none too happy, and found the place.
This is certainly one of the most unusual places I've ever seen a BSD conference. The first photo in the Church of San Pietro web page is taken from inside the abbey, just outside the room in which we had the conference. It's quite pretty, and I later discovered that the Green Michelin guide to Italy gives it two stars, the same as the cathedral in Firenze; I don't think it's quite that interesting, though.
Presented my own talk just before lunch. It's the first time I've done it, and the general impression was that I went into too much detail on the process of brewing. To be improved.
In the afternoon a talk by Guiseppe Maxia on MySQL performance. Interesting ideas; after the event had a talk with him before going out for dinner with nearly 20 people at the Osteria dell'artiste, not good food. Again far too late to bed.
| Sunday, 4 December 2005 | Perugia –> Milano | Images for 4 December 2005 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
The sun was shining this morning, and the view out of my hotel room looked better:
|
|||||||||||||||||||||||
At the conference, listened to a number of papers in Italian, which did more for my Italian comprehension than for the papers themselves.
Lunch was in the middle of town again:
|
|||||||||||||||||||||||
We certainly got enough exercise here, especially with the hilly aspect of the town, which left me wondering what they did for water.
In the evening, off to the station to catch the train to Milano, somewhat hampered by the fact that we didn't know where it was. Arrived at the station just in time to run and get on the train—not a thing for which my baggage is suited—but didn't get the tickets. Fortunately they had been paid for on the web, so we didn't have too much difficulty.
I'm continuing to Frankfurt, so in Firenze tried to find information about onward trains. That was a non-starter—they took too long to handle people—but on exiting from the ticket office, I discovered that there were two trains to Milano, and none of our people to be seen. It wouldn't have worried me so much if I hadn't left my computer bag with Max; fortunately, got on this train just in time as well. Spent the time discussing the next BSDCon Europe with Max and FreeSBIE with Dario Freni.
Arrived in Milano round midnight. It's been snowing a lot lately—a number of people missed the conference because the roads were impassable—but we were still surprised how much snow there was in Milano:
|
|||||||||||||||||||||||
Milano Centrale station is an impressive edifice, but it doesn't have any lifts, and the trains arrive upstairs:
|
|||||||||||||||||||||||
Josette Garcia of O'Reilly was also looking for a hotel, so first went down with Max looking for assistance. Didn't find that, but we got back to Josette just in time to see somebody snatch her bag with all the takings from the weekend and run off—in our direction. Grabbed the bag and chased him down the stairs (the left hand ones in the photo above). Didn't catch him, but I did have the satisfaction of seeing him fall down and roll down the stairs a couple of times.
Off into town looking for a hotel, and was lucky enough to find one just round the corner: the Amadeus, who gave us two rooms for €60 each per night, presumably way below the normal cost of the rooms (the web site gives prices round € 108). It's about the cheapest we've paid anywhere in Italy. Got a taxi to take our voluminous luggage round. It was worth the €10 I paid him to hear him curse and swear about having to do such a short journey. Arrived in the hotel with Josette giggling like a schoolgirl, and the receptionist apparently thought she was drunk. That wasn't the case, and we couldn't even get a beer to drink, so had water instead before a very late night.
| Monday, 5 December 2005 | Milano –> Frankfurt –> | Images for 5 December 2005 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Off to the station first thing in the morning and found an incredible disorganization. In the eastern ticket office they make reservations for trains that require them; in the western office they sell tickets for trains that don't. How do you know which? Trial and (almost inevitably) error. I chose East first; I don't know if it was right, but the ticket agent's computer (Microsoft, of course) froze up while she was checking, and she sent me to the Western office. There I was told that there was no way to get to Frankfurt/Main Airport, and that the only train to Frankfurt/Main main railway station left at 21:30.
On a second attempt she discovered what I had already guessed, that there was a train every hour; however, the number of changes varied from hour to hour. With some difficulty, including a supervisor trying to send me to several places with similar sounding names (shades of last Friday) was finally told that I could leave at 11:25 for Basel Badischer Bahnhof and there change for Frankfurt airport. Got a ticket with no train details, as I discovered only later.
Picked up my baggage and off on the train, which took fully 5½ hours of fortunately relatively pleasant travel to get to Basel SBB, where it terminated, 5 km before Badischer Bahnhof. That wasn't a big deal: a ticket inspector had already told me that I could catch the train in Basel SBB as well. That connection was comfortable; it's good to be back where I understand the system and the language better, and things seem generally better organized.
One of the interesting things about changing trains was that I had to go up a long, wide staircase to get the the new train. There was (apparently) no escalator. But a muscular female Swiss SBB employee came and carried my bag for me, apparently because I looked so old (57).
First ticket check in Germany brought the next (slight) surprise: the train they booked on me (Deutsche Bahn ICE) costs more. They had told me to take that train, but sold me the wrong ticket, so I had to pay € 10 more. Not a big deal, but somehow indicative of the general incompetence.
Arrived in Frankfurt Flughafen with no problems apart from the usual terrible signs, which seem to be deliberately placed to mislead. As I was getting into the bus to terminal 2, two Indians came along in a hell of a hurry. I told them not to worry: it was 20:15, and the bus wasn't scheduled to leave until 20:20. They said that their plane was scheduled to leave at 20:15, and would I please ask the driver to leave immediately?
Did that, and to my surprise the driver agreed. I had some time, so went with them when we got to the terminal. Of course, the plane had left, but they were given a booking for tomorrow and a discounted airport hotel (€ 55 for a double room). I suppose it's an everyday occurrence, but it's a first for me. It's good to remember the problems that people have.
Then checked the toilets (see my report from last year) and found that they were, indeed, all functional. Through security, not once but twice in rapid succession—even the security guards gave the impression that something was wrong with the system—only to discover that the “Qantas” lounge was really Delta (so they had plenty of spirits but no wine), and that I would have to go through security a third time after leaving it. What a pain.
Left the blocks dead on time and headed for the runway, where the capitan announced that they had to return for repair: one of the flaps systems wasn't working. That proved to be a defective switch, and they replaced it in a little over an hour.
| Tuesday, 6 December 2005 | –> Singapore –> | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Another day spent in a plane. What can I say? Economy class, last row in a cabin (so the seats reclined even less than normal), and screaming babies until Singapore. Got marginally enough sleep.
| Wednesday, 7 December 2005 | –> Sydney –> Adelaide | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Despite the delay on Monday night, got into Sydney only 20 minutes late and had no difficulty getting my connection to Adelaide; in fact, it meant that my total flight time from leaving Frankfurt to arriving in Adelaide was less than 24 hours, which I think is the first time I've managed that.
Called Yvonne from Sydney and discovered that we had had a power surge last night that had caused significant damage. On return, gradually found:
Got most things up and running, though the fuse in the printer is hard to get to, so I may take that down for repair.
Apart from that, spent the day in the usual way after returning from a trip, addressing the thousands of mail messages waiting for me.
| Thursday, 8 December 2005 | Echunga | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Spent most of the day catching up, but I also had a deadline to attend to: I need a report in by Monday. While working on that, discovered that the power surge the other night did more harm than I thought: the smell in the Mike Smith Memorial room proved to be the demise of two computer power supplies that weren't behind a UPS; they weren't even powered on at the time of the failure. On the positive side, it looks as if ETSA may pay for the damage.
As a result, once more harnessed my veteran HP LaserJet 6MP to echunga. Printing went at a snail's pace. Finally I discovered the message:
Interrupt storm detected on "irq7: lpt0"; throttling interrupt source
It proved to be a new interrupt throttling feature in the system: the sysctl variable hw.intr_storm_threshold sets the maximum number of interrupts per second on any interrupt level. The default value is 500, woefully inadequate for a PostScript printer on a parallel port, which can generate over 100,000 interrupts a second. Fixed that:
=== root@echunga (/dev/ttyp1) ~ 130 -> sysctl hw.intr_storm_thresholdhw.intr_storm_threshold: 500=== root@echunga (/dev/ttyp1) ~ 131 -> sysctl hw.intr_storm_threshold=200000hw.intr_storm_threshold: 500 -> 200000
Unfortunately, this effectively disables the interrupt storm detection system-wide. The values should be per interrupt.
In the afternoon to another board meeting of the ICT Council for South Australia. There was an annual general meeting and election of a new board while I was away, but there are few changes in the composition. Looks like we're finally getting organized.
| Friday, 9 December 2005 | Echunga | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Haven't really got back to normal yet, but I have a deadline, so my benchmark mail count is still much higher than normal.
I'm back to source code analysis at the moment, not for the first time this year. I still don't have a good tool; I wonder if I should waste some time writing something of my own.
In the afternoon, checked my credit card statement and discovered that AVIS had deducted the princely sum of AUD 1,057.49 from my credit card, about € 675. The agreed sum was € 112.82, about AUD 177. Called up AVIS and discovered:
I was unhappy with AVIS before this incident; now I am furious. And this is the company that “tries harder”. They'll have to try a whole lot harder than that if they ever expect me to do business with them again.
| Saturday, 10 December 2005 | Echunga | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Still chasing my deadline today, so spent most of the day doing that. Off in the afternoon on a wild goose chase looking for a Chinese cabbage; they must be out of season, and we finally returned with none. I've been driving past Hahndorf venison for ever with the intent to stop one day, and today was the day. Didn't buy anything there either—the main question was to find out what they had—so now we'll have to spend some time reading cookbooks.
| Sunday, 11 December 2005 | Echunga | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Finally a day I can go riding again! It's been over two months since the last time. Unfortunately, the weather wasn't the most ideal: the temperatures reached 35° today, and though we left early in the morning, the temperature was still nearly 30°. To add to the fun, the flies are particularly bad this year, probably due to the mild winter. Didn't stay out long.
More work on my code analysis, and more or less broke the back of it. It's one thing to understand code if you know in advance what it does; reading unknown code without any documentation doesn't make it any easier.
While I was in Helsinki, Monty showed me a well-worn copy of Bonniers Kokbok, which I found interesting, so when I was in Stockholm, I bought a copy. Although it's obviously a Swedish cookbook, it has a more universal coverage, and the recipes look good (though sometimes the otherwise good photos look very different from the recipes, a thing I have noted before). Today made a couple of fish dishes which were really quite good. To be recommended.
| Monday, 12 December 2005 | Echunga | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
On with my code analysis today, and finally got it finished. It's great to be back home and have everything you need in place.
Yvonne off in the afternoon to catch up with the damage done last Tuesday. In the process, discovered that there was more wrong with the printer than the fuse: at least the power supply board is dead. Make that two boards and it may be cheaper to buy a new printer. Grr.
| Tuesday, 13 December 2005 | Echunga | Images for 13 December 2005 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Shane Adcock and his mate came along today to install the new UPS and put the cables for the projector and loudspeakers in the lounge room ceiling. After my experience in August, I decided to let somebody else do the latter.
One way and another, it took the whole day. It's nice to have the security of a completely protected office, though; got the proof of that with the estimate for repairing the Brother colour laser printer. It needs at least a new power supply board, and only then will we know whether any further damage has been done. The power supply costs about $850; a new printer costs $708. That makes it pretty clear which to choose. The printer also includes four half-full toner cartridges and an OPC belt. Full toner cartridges cost $163, and the OPC belt costs $344, so with half-full cartridges the consumables add up to $670, leaving a cost of only $38 for the printer. What crazy prices!
| Wednesday, 14 December 2005 | Echunga | Images for 14 December 2005 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Finally caught up on the brewing that I had planned to do over the weekend, and as a result didn't do very much else.
Yvonne came back with a new Brother HL-2700CN colour laser printer. I was wrong in my assessment yesterday: I had forgotten that the printer also includes a fuser unit. I have no idea what they cost, and they're not listed by my supplier, but I'd assume they would have to cost more than $38, so basically the consumables supplied with the printer are worth more than the cost of the entire machine.
This machine has newer firmware than the old one, and one of the things that it does is to send email on various conditions. Unfortunately, the email upsets SpamAssassin:
Content analysis details: (3.2 points, 1.5 required)
pts rule name description
---- ---------------------- --------------------------------------------------
1.0 NO_REAL_NAME From: does not include a real name
0.2 NORMAL_HTTP_TO_IP URI: Uses a dotted-decimal IP address in URL
0.0 BAYES_50 BODY: Bayesian spam probability is 40 to 60%
[score: 0.4990]
0.2 MIME_BASE64_NO_NAME RAW: base64 attachment does not have a file name
1.9 MIME_BASE64_TEXT RAW: Message text disguised using base64 encoding
0.0 UPPERCASE_25_50 message body is 25-50% uppercase
Ended up having to put in a procmail rule to catch it.
| Thursday, 15 December 2005 | Echunga | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Into town today to talk with my tax people, and after that met up with David Logan of HP. David's a MySQL afficionado, and he's in the process of starting up a user group in Adelaide. It'll be interesting to hear how that pans out.
After that picked up the monitor that burnt out last week. The repairman said that there was nothing wrong with it, and his demonstration bore that out. How can a monitor still work after pouring out thick black smoke?
Lesson of the day: don't put digital thermometer probes into a closed barbecue. It seems that the cables can't take the heat, and now I have a dead probe.
| Friday, 16 December 2005 | Echunga | Images for 16 December 2005 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
More work on the table space API today, and finally got something worth sending to Monty.
While at the tax advisors yesterday, Nick recommended the Mind Your Own Business accounting package to me, and I ordered one from him. Today I discovered I can download a trial copy. I should have done that first: despite what Nick said, it's full of Microsoft GUI horrors. For example, it comes with a default list of accounts and a tiny window to edit them in:
|
|
||||||||
Now admittedly the image on this page is smaller than the original (click on it for the original), but on my main screen (wantadilla:0.1, 1600x1200) it's hard to read, and on my higher-resolution screen (wantadilla:0.0, 2048x1536) it's almost illegible. MYOB does not seem to offer any way to change the text size. More importantly. though, the only way to edit this overly verbose list (for example, to remove unneeded entries) is to click on each one individually, then move the mouse, click on delete and repeat. To make matters worse, I don't even get transparent scrolling, so effectively scrolling doesn't work.
Nick thought I might have problems with other issues—dual entry bookkeeping for example—but it seems that people are so used to the limitations of the Microsoft GUI that it no longer upsets them.
More phone calls in the afternoon, with Antony Curtis. It looks like a lot of us came away from Athens with new ideas about transaction processing, and Antony's and mine are similar. It's difficult to see how similar, since we're looking at them from a different perspective, but he left me with a lot to think about. There are some interesting concepts here.
| Saturday, 17 December 2005 | Echunga –> Olivaylle | Images for 17 December 2005 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
The weather this spring and summer has been really strange: it's been very wet, and there are currently no signs of it letting up. This morning we found that a storm had broken off some branches in the driveway:
|
|||||||||||||||||||||||
Despite that, off to Olivaylle to take a horse to Christiane Yeardley. The wind didn't make it any easier, and in the Wimmera (between the Little Desert and Big Desert national parks) got the car covered in mud:
|
|||||||||||||||||||||||
We had intended to go riding in the afternoon, but it was too windy, so didn't do much.
| Sunday, 18 December 2005 | Olivaylle –> Echunga | Images for 18 December 2005 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Slow day today. In the morning, Chris showed us around the new olive pressing plant:
|
|||||||||||||||||||||||
When they started this venture, Jorge had intended to build the biggest olive plantation in the world. It seems that he's given up on that idea, so now he's going for high quality instead. One implication: eye-catching glass bottles for the oil, apparently made in Tuscany:
|
|||||||||||||||||||||||
The wind had let up a little, so after that off on a short ride round the olive groves. Chris was riding Mucho (Magia's colt), whom we had called El Mago Negro), and Yvonne was riding Harry, both of whom were less than easy to handle. I had Sue Macintyre's Peruvian Paso mare Rosita, who is willing but inexperienced. I now recall why I brought Darah with me last time.
I originally wrote that Rosita was a Paso Fino; that's because of her gait. I've already had enough grief because of that mistake.
Back home against the wind. That alone must have cost us $20 in petrol.
| Monday, 19 December 2005 | Echunga | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
As usual on Mondays, spent most of the day catching up mail from the weekend and on administrative things. It's made easier by the fact that all my hard deadlines have been met, so now I need to prioritize the rest of my to-do list. Looks like time to examine some bugs.
David from “the broadband centre” called up to tell me that my exchange has now been “broadband enabled”, and that I should register with Telstra. I asked him about the application for ADSL that I made back in March, but he didn't know anything about that: “We're just a marketing company”.
I wonder if Telstra will ever rediscover this order.
When we were in Olivaylle, we borrowed some DVDs from Chris. Tonight watched one of them, Dr Strangelove, not for the first time. The link claims it's “the hot-line suspense comedy”, but I'd put it down as one of the blackest satires I've ever seen. What really worries me, though, is that it's over 40 years old, but it's just as convincing today. The military really doesn't seem to have learnt anything.
| Tuesday, 20 December 2005 | Echunga | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
More discussions with Antony today about alternative methods of transaction logging, and we're coming up with some interesting ideas. Basically, we're getting away from the idea that the table on disk is the definitive version of the table in an active database, and we're using it just as a backup for the in-memory view. Until it's committed, rows touched by a transaction only exist in the redo log buffers, and the engine knows how to find them. A transaction is committed when the redo log hits disk, and any other tidy-ups can happen later. This also means that recovery is almost the same thing as normal operation. None of this is revolutionary, of course, but the overall view makes it look very simple.
Also spent some time trying to get MySQL to build better under FreeBSD, with some (but only some) success. This build process is more fragile than I like.
Into town for an ADUUG lunch, which was only barely worth the trouble: Sue didn't show up, and Daniel and Fiona O'Connor will be here next week anyway, leaving just Bob.
Then to see Ramana Rao, my financial advisor, for another only barely worthwhile visit; as I feared, things weren't ready, so we just did a bit of talking. I'll have to go back again in the New Year.
| Wednesday, 21 December 2005 | Echunga | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Back to trying to build MySQL under FreeBSD today, and it's more difficult than I expected. Tried to install the necessary tools on our FreeBSD machine in Uppsala only to discover that my password has changed. Who knows when? I normally log in via ssh, so I only need a password when I run sudo. Put in a request for password reset and moved on to the next problem, which required building on Linux. That failed too (eventually; my systems are slow).
In the meantime spent some time looking at the Sun machines that Yvonne had brought back last week. There's a 10000 chassis with what proved to be only 2 CPU boards. It didn't have a display card on it, and when I put one in, the POST didn't talk to it. I wonder how to get that to work.
Planning (far too late) for the linux.conf.au 2006, which this year is in Dunedin. Booking flights is a pain!
| Thursday, 22 December 2005 | Echunga | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
I had planned to finally close a couple of bugs today. It wasn't to be. The whole day was a series of interruptions. It started in the morning when I got in and found that my password had not yet been reset. Spent some time trying to first access the problem report, which failed with password problems. Attempts to reset it via the web interface didn't seem to work, but somehow I finally got to it by means that I didn't understand, only to find that I wasn't able to update the severity.
Elliot suggested calling the Cupertino help desk, so decided to first—finally—set up my Sipura SPA-3000 ATA to dial to our internal VoIP network. That wasn't done by the time I had to call up somebody for an interview, and after that was over I discovered that the Cupertino desk was closed.
AVIS has still not responded to my complaint about the rental at the beginning of the month. In fact, I have not received any communication of any kind from them, neither a receipt for what they have charged me (which should have come automatically), nor a confirmation of my complaint, let alone a resolution (which had been promised me by now). I find it incredible that an international company can be so incompetent.
Contacted them in Sydney to ask for a fax number, and Diane gave me the number 03 9353 9017. I asked her if it was in Melbourne and she said no, in Sydney. I suggested that maybe she meant 02 9353 9017, which she confirmed. This incompetence is unbelievable. She was unable to give me the number for Italy, however, and put me through to a different department, where I spoke to Sarah.
After a long time she came back with the number: 9353 9119. I asked her if that included the Italian area code. No, the number was “here” (in other words, in Sydney; why do people in Sydney give numbers without area codes, as if there were no other phone areas in Australia?). She couldn't explain why the number was different from the one Diane gave me, and she couldn't give me the number in Italy.
I asked to be connected to her supervisor, and was connected to Kelly, who was also unable to give me the number in Italy, and who gave me yet another fax number, 1 800 123 703. She, too, was unable to explain the purposes of the various numbers. Sent a letter to each number, asking for immediate response, which of course was not forthcoming. It looks as if I'm going to have to take legal action.
Then spent some time trying to book a flight to Dunedin, which took me another two hours, not the least caused by broken web sites. The best bet was a “local” carrier, Freedomair, whose web site looked relatively unbroken for an airline web site:
|
||||||||||||||||||||||
But after filling out the flight dates, what next? Where the Search button? It's hidden behind the “Route Guide” and “Ask a question” images. Arjen helped me there: with the settings View –> Page Style –> No Style (under firefox) the display changes to this:
|
||||||||||||||||||||||
Getting seating was still a challenge: it seems that the URL for changing to the return flight was broken, and the only way to get to it was to tab through the URLs and watch what appeared at the bottom of the screen. Again a trick that Arjen taught me.
After that returned to trying to get my Sipura dialing strings sorted out. They seemed to work, but I didn't establish connection, as ethereal told me. Discovered that the adaptor configuration allows logging of different debug information, and soon discovered:
Dec 22 17:48:56 homephone SYSCFG:WARMBOOT 26 84 Dec 22 17:48:57 homephone fu:0:3b10, 0084 0086 0001 Dec 22 17:48:57 homephone fu:0:3b3d, 0038 03e4 0001 Dec 22 17:48:59 homephone subnet mask: 255.255.255.0 Dec 22 17:48:59 homephone gateway ip: 192.109.197.137 Dec 22 17:48:59 homephone dns servers: Dec 22 17:48:59 homephone 192.109.197.137 Dec 22 17:48:59 homephone Dec 22 17:48:59 homephone RSE_DEBUG: reference domain:mysql.com Dec 22 17:49:00 homephone TP Parser error: 14 Dec 22 17:49:00 homephone [0]SIP:ICMP Error -1 (d573a21d:5060, 1) Dec 22 17:49:00 homephone RSE_DEBUG: getting alternate from domain:mysql.com Dec 22 17:49:00 homephone IDBG: st-0 Dec 22 17:49:02 homephone RSE_DEBUG: getting alternate from domain:mysql.com Dec 22 17:49:02 homephone [0]RegFail. Retry in 30 Dec 22 17:49:02 homephone RSE_DEBUG: unref domain, mysql.com Dec 22 17:49:02 homephone RSE_DEBUG: last unref for domain mysql.com
So what does that mean, particularly the last two lines? No idea. And of course there's no documentation.
Dec 22 17:49:02 homephone fs:10389:10416:65536 Dec 22 17:49:02 homephone fbr:0:0000:0000:03b3d:0004:0003:2.0.13(GWg) Dec 22 17:49:02 homephone fhs:01:0:0000:upg:app:0:2.0.13(GWg) Dec 22 17:49:02 homephone fhs:02:0:0001:upg:app:1:2.0.13(GWg) Dec 22 17:49:02 homephone fhs:03:0:0002:upg:app:2:2.0.13(GWg) Dec 22 17:49:02 homephone fu:0:3b5d, 0003 0001 Dec 22 17:49:09 homephone TP Parser error: 14 Dec 22 17:49:29 homephone last message repeated 2 times
Another confusing one.
Dec 22 17:49:32 homephone RSE_DEBUG: reference domain:mysql.com Dec 22 17:49:32 homephone [0]SIP:ICMP Error -1 (d573a21d:5060, 1)
Yet another confusing message.
Dec 22 17:50:09 homephone Discard RSP w/o trnsac Dec 22 17:50:40 homephone last message repeated 3 times Dec 22 17:51:50 homephone last message repeated 7 times
This one was at least partially understandable: it came every 10 seconds, when an incoming UDP packet camed from sip.mysql.com.
More interesting is what happened when I tried to make a call:
Dec 22 17:52:47 homephone [0]Off Hook Dec 22 17:52:48 homephone Calling:4485@:0
This is, indeed, the number I was trying to call
Dec 22 17:52:48 homephone [0:0]AUD ALLOC CALL (port=16388) Dec 22 17:52:48 homephone [0:0]RTP Rx Up Dec 22 17:52:48 homephone RSE_DEBUG: reference domain: Dec 22 17:52:48 homephone RSE:GetServerAddrErr(,0)=-101 Dec 22 17:52:48 homephone TP:?Tx->0 Dec 22 17:52:48 homephone [0:0]AUD Rel Call Dec 22 17:52:48 homephone CC:Failed Dec 22 17:52:48 homephone RSE_DEBUG: unref domain, Dec 22 17:52:48 homephone RSE_DEBUG: last unref for domain Dec 22 17:52:48 homephone Sess Terminated Dec 22 17:52:50 homephone Discard RSP w/o trnsac Dec 22 17:52:51 homephone [0]On Hook
So what does all that mean? It looks as if I have DNS problems, but ethereal tells me that the system isn't trying any DNS requests. Also, I do have connectivity with sip.mysql.com. The messages don't make sense.
Fortunately I have a second Sipura machine, which I use to access Nodephone. Tried the same thing with that and discovered:
Dec 22 17:51:56 officephone subnet mask: 255.255.255.0 Dec 22 17:51:56 officephone gateway ip: 192.109.197.137 Dec 22 17:51:56 officephone dns servers: Dec 22 17:51:56 officephone 192.109.197.134 Dec 22 17:51:56 officephone 192.109.197.137 Dec 22 17:51:56 officephone Dec 22 17:51:58 officephone RSE_DEBUG: reference domain:sip.internode.on.net Dec 22 17:51:58 officephone fu:0:3be6, 03f6 0001 Dec 22 17:51:58 officephone [0]RegOK. NextReg in 59 (1) Dec 22 17:51:58 officephone IDBG: st-0 Dec 22 17:52:01 officephone fs:11313:11352:65536 Dec 22 17:52:01 officephone fbr:0:0000:0000:03baa:0006:0005:2.0.13(GWg) Dec 22 17:52:01 officephone fhs:01:0:0000:upg:app:0:2.0.13(GWg) Dec 22 17:52:01 officephone fhs:02:0:0001:upg:app:1:2.0.13(GWg) Dec 22 17:52:01 officephone fhs:03:0:0002:upg:app:2:2.0.13(GWg) Dec 22 17:52:01 officephone fhs:04:0:0003:upg:app:0:2.0.13(GWg) Dec 22 17:52:01 officephone fhs:05:0:0004:upg:app:1:2.0.13(GWg) Dec 22 17:52:01 officephone fhs:06:0:0005:upg:app:2:2.0.13(GWg) Dec 22 17:52:01 officephone fu:0:3bf8, 0003 0001 Dec 22 17:52:03 officephone RSE_DEBUG: unref domain, sip.internode.on.net Dec 22 17:52:03 officephone RSE_DEBUG: last unref for domain sip.internode.on.net
That looks surprisingly similar. What does it mean? I have no trouble accessing sip.internode.on.net.. It looks as if the “reference domain” messages are just plain wrong.
Making a call was instructive, though:
Dec 22 17:52:23 officephone [0]Off Hook Dec 22 17:52:26 officephone Calling:83888725@sip.internode.on.net:0 Dec 22 17:52:26 officephone [0:0]AUD ALLOC CALL (port=16396) Dec 22 17:52:26 officephone [0:0]RTP Rx Up Dec 22 17:52:26 officephone RSE_DEBUG: reference domain:sip.internode.on.net Dec 22 17:52:26 officephone RSE_DEBUG: reference domain:sip.internode.on.net Dec 22 17:52:27 officephone CC:CallProgress Dec 22 17:52:27 officephone [0:0]ENC INIT 18 Dec 22 17:52:27 officephone [0:0]RTP Tx Up (pt=18->cb028601:31594) Dec 22 17:52:27 officephone [0:0]RTCP Tx Up Dec 22 17:52:27 officephone [0]On Hook Dec 22 17:52:27 officephone [0:0]AUD Rel Call Dec 22 17:52:27 officephone RSE_DEBUG: reference domain:sip.internode.on.net Dec 22 17:52:27 officephone DLG Terminated Dec 22 17:52:30 homephone Discard RSP w/o trnsac Dec 22 17:52:32 officephone RSE_DEBUG: unref domain, sip.internode.on.net
In other words, it looks as if homephone doesn't know the system to contact. Not knowing the structure of the messages makes it almost impossible to debug. Still, I'm closer now.
| Friday, 23 December 2005 | Echunga | Images for 23 December 2005 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
AVIS is really getting on my nerves. Sent them another strongly worded letter today, and to my surprise got a reply (from Chris Adams, in New Zealand). He told me, in variance from what I had heard before, that it could take up to a month to get resolution, and that I should thus expect a response some time before 9 January. He also hadn't received yesterday's fax (sent to the same fax numbers), so sent it on to him again.
Our plans for linux.conf.au 2006 are still not over. I have my flights, but we spent some time discussing Brian Aker's itinerary. It looks as if he'll have time to come to Adelaide after the conference and open the user group that David Logan has been planning to start up. Spent far too much time discussing that.
More work on the Sipura SPA-3000 ATA, but not enough to work out what is going wrong. It's still extremely convenient to have the debug output from a functional machine to compare.
More work on my bugs, but didn't get anything finished. What a pain!
| Saturday, 24 December 2005 | Echunga | Images for 24 December 2005 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Up early this morning to prepare for this evening's dinner: relatively traditional turkey with accompaniments. Spent a bit of time documenting it for a change, and made pretty good progress:
|
|||||||||||||||||||||||
A far cry from the Turkey with whisky of a couple of years ago.
The weather is getting no better. The heavy winds continue, and the gum tree that lost a branch a couple of days ago has now lost everything:
|
|||||||||||||||||||||||
In mid-afternoon, Chris Yeardley called: she had been delayed, and wouldn't be there in time for dinner. Fortunately the turkey hadn't made it to the oven, so with a bit of grumbling changed the menu and had some curries from the deep freeze instead. We'll eat the turkey tomorrow.
| Sunday, 25 December 2005 | Echunga | Images for 25 December 2005 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Despite the weather forecast, the weather was not very good this morning. Out riding in Kuitpo forest anyway.
The rest of the day was dedicated to yesterday's Christmas dinner. Spent most of the afternoon preparing, somewhat animated by the fact that Chris has a new camera and took many photos:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
The dinner itself was a great success. I must write up the technical details; but I got everything just the way I wanted it.
| Monday, 26 December 2005 | Echunga | Images for 26 December 2005 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Today was the sixth annual Hacker's barbecue, and the weather was ideal. We had intended to do some multimedia work, but somehow that didn't happen; instead, Daniel O'Connor brought along various hardware projects he's been working on, including a USB interface to a Dallas 1820 thermometer probe:
|
|||||||||||||||||||||||
He's also been playing with Linksys ADSL routers:
|
|||||||||||||||||||||||
The result of that one was an unplanned outage of my DSL line:
|
|||||||||||||||||||||||
| Tuesday, 27 December 2005 | Echunga | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
We had planned to go riding with Michael Hickinbotham in his vineyards today, but it was clear pretty early that it would be too warm, so postponed. Instead started what promised to be a pleasantly quiet day.
That changed shortly after midday, when the weather changed. A sudden clap of thunder, and of course the power failed. As I had feared, the new UPS didn't like the generator, and after about 90 minutes running alternately on generator and battery, emptied the latter and powered down.
When power finally came back, not all the hardware did. This is becoming too common. By the end of the day, discovered that my ADSL modem, my Cisco 2900 switch and both Sipura SPA 3000s were dead:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
The second time in a month! This is completely unacceptable, and of course it meant that I was off the net. Got dialin information from Internode, but of course that doesn't route my network, so, for example, I can't get mail.
| Wednesday, 28 December 2005 | Echunga | Images for 28 December 2005 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Up early to organize the replacement network hardware. Things didn't go very smoothly: first, it turns out that my supplier is closed until next week, so there was no way to get a new switch. Then there were timing difficulties with getting the equipment from Internode. Fortunately, Daniel O'Connor had a spare switch that he was able to lend me, so Yana set off into town and returned shortly before 1 pm with a switch and an ADSL modem—and no Sipura VoIP ATAs! They had so completely forgotten to give them to her that they (apparently) removed them from on top of the box with the ADSL modem and didn't give them to her even when she asked if that was all.
She set off back into town while I tried to install the modem (a Linksys AG241). It wasn't easy last time, but I had expected it to be easier this time. The problems:
Like many other devices, it came with a RFC 1918 default address 192.168.1.1. The instructions say “Enter http://192.168.1.1 in the web browser's Address field.”
That's all fine and good, but it assumes that you have a route to the network. This step just Doesn't Work, not even with Microsoft. On MacOS X it does exactly the same thing as with FreeBSD: you need a route to it. So why don't more people have problems with this issue? Are they all running NAT with addresses in the 192.168.0.0/16 address space?
I think the real answer lies in the “PC mentality”: reboot, reboot, reboot. At the beginning, the instructions state: “Make sure that all the devices that you'll be working with are powered down, including your PCs and the Gateway.” Apart from the fact that I can't even begin to imagine such a concept, it would still only work if the first PC you turn on got an address via DHCP. Then it would work, of course, until you change your address: the modem includes a DHCP server that requires to be disabled.
After that, I still couldn't access the web configuration. The problems appear to be related to reverse DNS lookup. This is a chicken-and-egg problem: I couldn't get DNS working properly until the link is up, and I couldn't get the link up until I ran configuration. Fortunately there was a timeout, so after a couple of minutes I was able to set it up.
By the time I had that finished, Yana was back with the Sipura ATAs. Surprisingly, the first one was relatively easy to configure: they themselves use DHCP, so the method is roughly:
A whole day spent mopping up the mess, not to mention a lot of yesterday. I really need to stop this happening.
| Thursday, 29 December 2005 | Echunga | Images for 29 December 2005 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Daniel Demuth showed up today on his annual Australian holiday. Unfortunately it was too hot to go riding (well, I thought so; Yvonne and Daniel went anyway), so spent the day trying to get Asterisk running. I failed, greatly helped by the book that I had bought. Did learn a little, anyway.
While on the subject of VoIP, turned my attention to the second Sipura SPA-3000 ATA that Yana had brought back yesterday. The intention is to use it via Asterisk to access both POTS and multiple VoIP servers, but the immediate need is to connect to the MySQL internal servers and be able to place calls. Spent some time looking at what documentation I have, and finally got something working. It's unbelievable how bad this documentation is. The dial string I got from Internode reads:
(000S0<:@gw0>| 106S0<:@gw0>| 183[12]x.| 1[38]xxxxxxxxS0<:@gw0>| 13[1-9]xxxS0<:@gw0>| 1[9]xxxxxxxxS0<:@gw0>| 0[23478]xxxxxxxxS0| [2-9]xxxxxxxS0| 001xxxx.S5| xxx.<:@gw0>| *xx.| <#,:>x.<:@gw0>)
The meaning of these strings is all described in Internode's setup guide, but it doesn't correspond to the description in the Sipura documentation, which doesn't mention the syntax <:@gw0> at all. In particular, it seems that the box is capable of connecting to multiple VoIP services, but I still can't work out how to do it.
Cheese fondue in the evening, as if Daniel hadn't had enough already. He took some photos so that he could show people back home that there are really people here who eat fondue in the summer:
|
|||||||||||||||||||||||
| Friday, 30 December 2005 | Echunga | Images for 30 December 2005 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Quiet day, also very hot, over 37° even here in the hills. Spent most of it writing up my experiences with Asterisk and the Christmas cooking of last weekend. Also found a Sipura SPA 3000 FAQ which helps a little, though I'm gradually coming to the conclusion that there's a conspiracy to document VoIP badly. For example, the FAQ contains several answers to specific questions about dial strings (including one that suggests that I can, in fact, do everything I want without Asterisk), but no examples at all. I still don't know where to find a complete definition of the dial string syntax.
| Saturday, 31 December 2005 | Echunga | Images for 31 December 2005 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Another hot day outside, so didn't go out much. Instead continued with the VoIP stuff, and made some progress. One thing remains: all related documentation is really terrible. Went through all the available information on the Sipura SPA-3000 and got a little further. Even found a completely broken “tool” on the Sipura web site that purports to set up dial plan (click on images to enlarge):
|
||||||||||||||||||||||
I suppose it's forgiveable that the IP address really needs to be given as an IP address (and I'm not sure why it's needed anyway), and I suppose it's “modern” to have boxes that are only half displayed, like the Select box. But the real problem is that the tool doesn't do anything useful. You can only select a few predetermined networks:
|
||||||||||||||||||||||
Once you do this, the application crashes:
|
||||||||||||||||||||||
I thought of contacting the webmaster, but the address resolves as root@localhost. What a mess!
Then spent some more time looking at the manual, which did have some additional information:
Section 4.14.1, on page 60, refers to a total of 8 dial plans that the unit can apparently store. This is the only place I've seen these mentioned, and it doesn't tell you how to use them. I'm left wondering if they ever got implemented.
Section 4.14.11, on page 67, finally mentions the <:@gw0> syntax, though of course it doesn't explain it completely. Still, it was enough for me to be able to set up a second gateway (to Nodephone) on the unit that already talks to the MySQL server, and it really did talk to the server. But despite all attempts, the messages were sent with my MySQL address as the originator:
INVITE sip:83888250@sip.internode.on.net SIP/2.0.
Via: SIP/2.0/UDP 192.109.197.162:5060;branch=z9hG4bK-b76644c4.
From: Greg Lehey <sip:4484@sip.mysql.com>;tag=b2ae18c845787224o0.
To: <sip:83888250@sip.internode.on.net>.
Call-ID: 187b4898-4e608494@192.109.197.162.
CSeq: 101 INVITE.
Max-Forwards: 70.
Contact: Greg Lehey <sip:4484@192.109.197.162:5060>.
Expires: 240.
User-Agent: Sipura/SPA3000-2.0.13(GWg).
Proxy-Require: Yes.
Content-Length: 430.
Allow: ACK, BYE, CANCEL, INFO, INVITE, NOTIFY, OPTIONS, REFER.
Supported: x-sipura.
Content-Type: application/sdp.
NodePhone didn't seem to like the result:
U +0.053726 203.2.134.1:5060 -> 192.109.197.162:5060 SIP/2.0 100 Trying. Via: SIP/2.0/UDP 192.109.197.162:5060;branch=z9hG4bK-b76644c4. From: Greg Lehey <sip:4484@sip.mysql.com>;tag=b2ae18c845787224o0. To: <sip:83888250@sip.internode.on.net>. Call-ID: 187b4898-4e608494@192.109.197.162. CSeq: 101 INVITE. U +0.010017 203.2.134.1:5060 -> 192.109.197.162:5060 SIP/2.0 604 Does not exist anywhere. Via: SIP/2.0/UDP 192.109.197.162:5060;branch=z9hG4bK-b76644c4. From: Greg Lehey <sip:4484@sip.mysql.com>;tag=b2ae18c845787224o0. To: <sip:83888250@sip.internode.on.net>;tag=SD2oeo799-477523675-1136072985619. Call-ID: 187b4898-4e608494@192.109.197.162. CSeq: 101 INVITE. Content-Length: 0.
One thing that this testing did reveal is that last Tuesday's storm claimed one more victim: one of the ringers on my two-line Panasonic phone no longer works. It's becoming more and more clear that the damage must have happened on the phone line this time.
Gave up on the Sipuras and installed a couple of soft phones from the FreeBSD Ports Collection, and even discovered that Linphone has a command line tool that isn't quite as painful as these silly windows. It's not exactly user friendly either, though:
=== root@eucla (/dev/ttyp1) ~ 74 -> linphonec call 0871270137
ERROR: bad arguments
usage: linphonec [-c file] [-s sipaddr] [-a] [-d level -l logfile]
linphonec -v
-c file specify path of configuration file.
-d level be verbose. 0 is no output. 6 is all output
-l logfile specify the log file for your SIP phone
-s sipaddress specify the sip call to do at startup
-a enable auto answering for incoming calls
-t enable auto terminate on process end
-v or --version display version and exits.
In other words, the command line syntax for calling is different from when the program is running in interactive mode. This syntax works once it's running:
=== root@eucla (/dev/ttyp1) ~ 75 -> linphonec
Ready.
linphonec> call 0871270137
Warning: Could not parse given sip address. A sip url usually looks like sip:user@domain
linphonec> ^P^A^A^C
And it doesn't have any command line editing.
=== root@eucla (/dev/ttyp1) ~ 76 -> linphonec
Ready.
linphonec> call sip:0871270137@sip.internode.on.net
Contacting sip:0871270137@sip.internode.on.net
linphonec> terminate
Communication ended.
linphonec> Request Cancelled.
Using the (different) command line syntax isn't all plain sailing, either:
=== root@eucla (/dev/ttyp1) ~ 78 -> linphonec -s sip:0871270137@sip.internode.on.net
Ready.
linphonec> Contacting sip:0871270137@sip.internode.on.net
^C
Stopping the process doesn't terminate the call. I next tried this:
=== root@eucla (/dev/ttyp1) ~ 79 -> linphonec -t
Ready.
linphonec> terminate
linphonec>
That didn't terminate the call either. It looks as if you can't under those circumstances: I had to answer the phone to terminate the call. And you have to type in full length URLs, and you have to explicitly terminate a call (wouldn't t be enough?). Finally, you need to use quit to exit the program; ^D causes it to loop on the command prompt. Still, with a bit of work (and hopefully some kind of registration) it's just what I need for when I'm travelling.
Do you have a comment about something I have written? This is a diary, not a “blog”, and there is deliberately no provision for directly adding comments. It's also not a vehicle for third-party content. But I welcome feedback and try to reply to all messages I receive. See the diary overview for more details. If you do send me a message relating to something I have written, please indicate whether you'd prefer me not to mention your name. Otherwise I'll assume that it's OK to do so.
| Top of page | Previous month | Greg's home page | Today's diary entry | Next month | Greg's photos | Copyright information |